Paper | Project Page | Video
ReLoBlur, the first real-world local motion deblurring dataset, which is captured by a synchronized beam-splitting photographing system. It consists of 2,405 blurred images with the size of 2,152×1,436 that are divided into 2,010 training images and 395 test images. For efficient training and testing, we also provide the resized version of ReLoBlur Dataset with the size of 538x359. ReLoBlur includes but is not limited to indoor and outdoor scenes of pedestrians, vehicles, parents and children, pets, balls, plants and furniture.
Baidu Netdisk | Google Drive | Number | Description | |
---|---|---|---|---|
ReLoBlur test | [link]() | link | 395 | We provide 395 pairs of testing images. |
ReLoBlur train | [link]() | link | 2,010 | We provide 2,010 pairs of training images. |
Resized | [link]() | link | 2,405 | We provide a resized version of ReLoBlur dataset. We resized ReLoBlur by BICUBIC algorithm. |
Local Blur Mask | [link]() | link | 2,405 | We provide a resized version of ReLoBlur dataset. We resized ReLoBlur by BICUBIC algorithm. |
Important: ReLoBlur dataset can be only used for academic purposes!
Based on ReLoBlur, we propose a Local Blur-Aware Gated network (LBAG) and several local blur-aware techniques to bridge the gap between global and local deblurring. LBAG detects blurred regions and restores locally blurred images simultaneously.
Before running LBAG, please install the environment on Linux:
pip install -U pip
pip install -r requirements.txt
Put the training data and testing data under
LBAG/data/dataset/
and put the masks under
LBAG/data/
If you train or infer with your own data, the data structure should be like:
├── dataset
├── train
├── s01
├── 00
├── 00_sharp.png
├── 00_blur.png
├── ...
├── ...
├── s02
├── 00
├── ...
├── ...
├── test
├── scene1
├── 00
├── 01
├── ...
├── scene2
├── 00
├── 01
├── ...
├── masks
├── train
├── s01
├── 00
├── 00.png
├── 01.png
├── ...
├── 01
├── ...
├── s02
├── 00
├── ...
├── ...
├── test
├── scene1
├── 00
├── 01
├── ...
├── scene2
├── 00
├── 01
├── ...
LBAG+ uses the pretrained model of MiMO_UNet, which could be downloaded here: https://github.com/chosj95/MIMO-UNet
And please put the pretrained model into
checkpoints/pretrained_mimounet/
You can infer LBAG by the pretrained model, which could be downloaded here:https://drive.google.com/drive/folders/1_T8Z7-T7i7BsaJhpq0rCNWLT65N_SslU?usp=sharing
You can evaluate the LBAG by using:
CUDA_VISIBLE_DEVICES=0 python LBAG_trainer.py --config configs/LBAG.yaml --exp_name 0223_munet_gate01_1 --reset --infer
Training with single GPU
To train a model with your own data/model, you can edit the configs/LBAG.yaml
and run the following codes:
CUDA_VISIBLE_DEVICES=0 python LBAG_trainer.py --config configs/LBAG.yaml --exp_name 0223_munet_gate01_1 --reset
Cication
Data on Baidu Netdisk
...